Ontology Modelling Approach for
Personality Measurement based on
Social Media Activity
Lab. Social Computing and Big Data
School of Economics & Business, Telkom University
©2018
Andry Alamsyah, Muhammad Rizqy, Darin Dindi
Changes in Behavior
▸ Post daily routines
▸ Share opinion, thoughts, and
information
▸ Communication
Activities on social media is stored as a digital trace
It provides us a new data resources
User also unconsciously reveal their personality through their writing and
linguistic text usage
- G. Stemmler, and J. Wacker, “Personality, emotions and individual differences in physiological responses,” Journal of Biological Psychology,
vol. 84, pp. 541-551, July 2010.
Today’s Personality Measurement
▸ Via Interviews, Test, or
Questionnaire Suvey
▸ Most likely answer by what
is socially acceptable.
Thehrdepartemenet.ie. Advantages dan Disadvantages of Employee Personality Tests, 2 October 2010.
OCEAN: Personality Traits
CO A NE
OPENNES CONSCIENTIOUSNESS EXTRAVERSION AGREEABLENESS NEUROTICSM
Do they like new
experiences?
Do they prefer planning and
regularity?
Do they like to spend a
lot of time with others?
Do they put concern
of others interests?
Do they tend to
worry?
OCEAN classifies human personality traits into 5 dimensions
and supported in over 50 societies across 6 continents
NEO-PI-R metric
CO A NE
OPENNES CONSCIENTIOUSNESS EXTRAVERSION AGREEABLENESS NEUROTICSM
Fantasy,
Aesthetic,
Feelings, Actions,
Ideas, Values
Competence, Order,
Dutifulness, Achievement-
Striving, Self-Discipline,
Deliberation
Warmth,
Gregariousness,
Assertiveness, Activity-
Level, Excitement-
Seeking, Positive
Emotions
Trust,
Straightforwardness,
Compliance,
Tendermindedness,
Modesty, Altruism
Anxiety, Angry
Hostility,
Depression, Self-
Consciousness,
Impulsiveness,
Vulnerability
Ontology
Ontology represents the
relation among entity thus
forming domain knowledge
Forming Components of Ontology
Methodology
Ontology OCEANX
Experiment Process
Twitter
Conversational
Data
Preprocessing
Words
Collection
Ontology
Construction
Ontology
Model
Personality
Measurement
Twitter Conversational Data
We use 8 samples i.e. actors to examine the performance of ontology model.
Account Real Name
Raw
Data
Filtered
Data
@aMrazing Alexander Thian 299 101
@jokoanwar Joko Anwar 124 66
@ulil Ulil Abshar-Abdalla 543 210
@gm_gm Goenawan Mohamad 83 36
@adelladellaide Adellina Fitriyani 741 243
@popokman Wandy Ghani 801 204
@duniamanji Anji 709 239
@pandji Pandji Pragiwaksono 145 71
Total
3.445 1.170
Preprocessing
Case Folding
URL and
Username
Removal
Symbol and
Number Removal
Tokenizing
Stopwords
Synonym
Recognition and
Word
Generalization
Phrases Lookup
Words Collection
Collecting words that are able to reflect personality traits
from Twitter’s Universe.
▸i.e. colloquial expression, slang, and phrases.
Ontology Construction and Ontology Model
Text
(Instance)
Facet
(Sub-Class)
Personality
(Class)
Neuroticsm
Anxiety
ribet aduh
Vulnerability
bingung
Personality Measurement
Word Personality Score Percentage
Benci Neuroticism 1 50%
Rindu Extraversion 1 50%
Result (Ontology Model Construction)
Personality
(Class)
Facet (Sub Class) Example Word
(Instance)
Openness Fantasy ingin_coba, mau_lihat,
paling_enak_dimana,
Aesthetics humor_terbaik, kreatif,
menarik, mengaggumi
Feelings butuh_bimbingan,
penasaran, ada_apa,
ingin
Actions aku_baru_coba,
bisa_kita_baca,
sharing
Ideas benarkah,
pandangan_saya,
kok_bisa
Values bebas,
perlu_dibaca_juga,
saya_kira_masuk_akal
Neuroticism Anxiety aduh, alay, anjir, bete,
bohong, jijik, lebay,
enggak_asik, rumit,
susah, ya_ampun,
berlebihan,
astaghfirullah, astaga,
ckck, ngeri
Angry Hostility bacot, bangke, goblok,
bangsat, marah,
ndasmu, njing, nyinyir,
sebal, benci,
tersinggung
Depression huhu, kecewa,
khawatir, sedih, pucat
Self-Consciousness enggak_kuat,
enggak_suka, capek,
belum_pede, malu,
nyesel
Impulsiveness bully, tidak_yakin
Vulnerability chaos, terluka,
bingung, cengeng,
persoalan_pelik,
bingung, melelahkan,
jahat, ngambek, nangis
Measurement Result
@jokoanwar @ulil
Conclusion
▸ Measuring human personality based on social media
activity can be modelled using ontology approach.
▸ The components of ontology model are personality traits
as a class, facets of the personality as a subclass, and
the words that represent personality as an instance.
Benefits of Research
▸ Able to integrate with machine learning for automatic
classification.
▸ Bridge the gap between ontology and psychology research.
▸ Ontology model construction based on Bahasa Indonesia
for measuring human personality.
▸ Gives minimal effort to measure human personality.
▸ Gives cheaper and faster option to measure human
personality.
Suggestion
▸Enrich the model’s dictionary.
▸Giving weight to each word based on the phrase they are on.
Thank you!

Ontology Modelling Approach for Personality Measurement based on Social Media Activity

  • 1.
    Ontology Modelling Approachfor Personality Measurement based on Social Media Activity Lab. Social Computing and Big Data School of Economics & Business, Telkom University ©2018 Andry Alamsyah, Muhammad Rizqy, Darin Dindi
  • 2.
    Changes in Behavior ▸Post daily routines ▸ Share opinion, thoughts, and information ▸ Communication
  • 3.
    Activities on socialmedia is stored as a digital trace It provides us a new data resources User also unconsciously reveal their personality through their writing and linguistic text usage - G. Stemmler, and J. Wacker, “Personality, emotions and individual differences in physiological responses,” Journal of Biological Psychology, vol. 84, pp. 541-551, July 2010.
  • 4.
    Today’s Personality Measurement ▸Via Interviews, Test, or Questionnaire Suvey ▸ Most likely answer by what is socially acceptable. Thehrdepartemenet.ie. Advantages dan Disadvantages of Employee Personality Tests, 2 October 2010.
  • 5.
    OCEAN: Personality Traits COA NE OPENNES CONSCIENTIOUSNESS EXTRAVERSION AGREEABLENESS NEUROTICSM Do they like new experiences? Do they prefer planning and regularity? Do they like to spend a lot of time with others? Do they put concern of others interests? Do they tend to worry? OCEAN classifies human personality traits into 5 dimensions and supported in over 50 societies across 6 continents
  • 6.
    NEO-PI-R metric CO ANE OPENNES CONSCIENTIOUSNESS EXTRAVERSION AGREEABLENESS NEUROTICSM Fantasy, Aesthetic, Feelings, Actions, Ideas, Values Competence, Order, Dutifulness, Achievement- Striving, Self-Discipline, Deliberation Warmth, Gregariousness, Assertiveness, Activity- Level, Excitement- Seeking, Positive Emotions Trust, Straightforwardness, Compliance, Tendermindedness, Modesty, Altruism Anxiety, Angry Hostility, Depression, Self- Consciousness, Impulsiveness, Vulnerability
  • 7.
    Ontology Ontology represents the relationamong entity thus forming domain knowledge Forming Components of Ontology
  • 8.
  • 9.
  • 10.
    Twitter Conversational Data Weuse 8 samples i.e. actors to examine the performance of ontology model. Account Real Name Raw Data Filtered Data @aMrazing Alexander Thian 299 101 @jokoanwar Joko Anwar 124 66 @ulil Ulil Abshar-Abdalla 543 210 @gm_gm Goenawan Mohamad 83 36 @adelladellaide Adellina Fitriyani 741 243 @popokman Wandy Ghani 801 204 @duniamanji Anji 709 239 @pandji Pandji Pragiwaksono 145 71 Total 3.445 1.170
  • 11.
    Preprocessing Case Folding URL and Username Removal Symboland Number Removal Tokenizing Stopwords Synonym Recognition and Word Generalization Phrases Lookup
  • 12.
    Words Collection Collecting wordsthat are able to reflect personality traits from Twitter’s Universe. ▸i.e. colloquial expression, slang, and phrases.
  • 13.
    Ontology Construction andOntology Model Text (Instance) Facet (Sub-Class) Personality (Class) Neuroticsm Anxiety ribet aduh Vulnerability bingung
  • 14.
    Personality Measurement Word PersonalityScore Percentage Benci Neuroticism 1 50% Rindu Extraversion 1 50%
  • 15.
    Result (Ontology ModelConstruction) Personality (Class) Facet (Sub Class) Example Word (Instance) Openness Fantasy ingin_coba, mau_lihat, paling_enak_dimana, Aesthetics humor_terbaik, kreatif, menarik, mengaggumi Feelings butuh_bimbingan, penasaran, ada_apa, ingin Actions aku_baru_coba, bisa_kita_baca, sharing Ideas benarkah, pandangan_saya, kok_bisa Values bebas, perlu_dibaca_juga, saya_kira_masuk_akal Neuroticism Anxiety aduh, alay, anjir, bete, bohong, jijik, lebay, enggak_asik, rumit, susah, ya_ampun, berlebihan, astaghfirullah, astaga, ckck, ngeri Angry Hostility bacot, bangke, goblok, bangsat, marah, ndasmu, njing, nyinyir, sebal, benci, tersinggung Depression huhu, kecewa, khawatir, sedih, pucat Self-Consciousness enggak_kuat, enggak_suka, capek, belum_pede, malu, nyesel Impulsiveness bully, tidak_yakin Vulnerability chaos, terluka, bingung, cengeng, persoalan_pelik, bingung, melelahkan, jahat, ngambek, nangis
  • 16.
  • 17.
    Conclusion ▸ Measuring humanpersonality based on social media activity can be modelled using ontology approach. ▸ The components of ontology model are personality traits as a class, facets of the personality as a subclass, and the words that represent personality as an instance.
  • 18.
    Benefits of Research ▸Able to integrate with machine learning for automatic classification. ▸ Bridge the gap between ontology and psychology research. ▸ Ontology model construction based on Bahasa Indonesia for measuring human personality. ▸ Gives minimal effort to measure human personality. ▸ Gives cheaper and faster option to measure human personality.
  • 19.
    Suggestion ▸Enrich the model’sdictionary. ▸Giving weight to each word based on the phrase they are on.
  • 20.